US 20060178950 A1
An inventive procurement system includes a requisition system for special item purchases that are not found when searching a catalog database. Using the requisition system the buyer is provided with the desired item while the information associated with the requisition is selectively added to a rule-based knowledge base as well as to an item database. In a preferred embodiment a common language generator has been used to normalize free form data using pre-determined rules to place the data into a class/attribute/value relationship. By having the item requisition follow the same relationship, the pre-determined rules may be updated and the free-form data associated with the item properly coded for inclusion into an item database.
21. A procurement system for sourcing of a buyer-specified desired item, said system comprising:
a catalog comprising a plurality of electronic catalog databases of catalog items, each said item being uniquely identified using a common parametric hierarchy with respect to normalized class relationships, attribute relationships and value relationships, said relationships including a source for each said item;
a catalog database selection interface being used by a buyer to select a plurality of electronic catalog databases of the catalog to be searched and to specify a cascading search of the selected databases to be searched for a desired item;
a hot list of previously purchased items identified by class relationships, attribute relationships and value relationships;
a search procedure based on pre-determined criteria for matching the relationships of the hot list and the catalog items to buyer-specified class relationships, attribute relationships and value relationships for the desired item in the selected databases of the specified cascading search,
wherein matching classes, attributes and values are captured in a special requisition whenever the item is not found in the hot list and not found in any of the selected electronic catalogs, said special requisition being used to select at least one potential source of the desired item.
22. The system of
at least two electronic catalog databases each having a different parametric hierarchy of class relationships, attribute relationship and value relationships; and
at least one exception list between a pair of the at least two electronic catalog databases, said exception list being used by said search procedure to match a normalized class relationship, attribute relationship or value relationship of a first electronic database of the pair with a corresponding class relationship, attribute relationship and value relationship in the other electronic database of the pair.
23. The system of
a standard buyer-specific electronic catalog database;
at least one intermediate electronic catalog database based on a topic selected from the group consisting of geography and subject matter;
a cumulative global electronic catalog database including all items tracked by a back office, said tracked items including items not included in any other electronic catalog database of the catalog.
24. The system of
25. The system of
26. The system of
27. The system of
This application is a continuation in part of U.S. application Ser. No. 09/348,693, filed on Jul. 7, 1999, the contents of which are incorporated in their entirety. The present invention relates in general to a procurement system. More specifically, the invention relates to a procurement system where a user completes a structured requisition form based predominantly on preexisting normalized relationships to order an item not available in a catalog database. The form is used to specify the desired item and to selectively create updated normalized relationships for use in identifying the new item so that it and similar items may be placed in the catalog database when free form data is examined using the normalized relationships.
Procurement systems are well known. A buyer searches a catalog or a catalog database, locates material of interest, and places an order. A supplier then fulfills the order. However, special orders greatly complicate and slow down the entire procurement process as well as greatly increasing procurement expense. In practice, a buyer who cannot find an item located in a catalog must fill out a special requisition, also known as a special order request. However, no order is even placed until a potential supplier and related cost and delivery information is provided to the buyer for consideration.
The requisition is then sent to a procurement fulfillment organization. A fulfillment specialist reviews the special requisition and manually determines if there are any potential suppliers that can fulfill the special requisition. Potential suppliers are then contacted and they respond accordingly. Each of the suppliers has their own way of describing the items it carries. Thus, the fulfillment specialist must manually review each supplier proposal and determine which ones appear to be most favorable to the buyer.
Next, the buyer must receive the best proposals from the fulfillment specialist, determine which one appears to be the most appropriate for that organization, and manually go through a special order process for the special requisition to actually be fulfilled by a supplier. Worse yet, when the special order is finally fulfilled, all of the special effort that went into manually reviewing the requisition, manually determining potential suppliers, manually locating the best suppliers, and then actually fulfilling the special order is lost. Thus, the entire process must be re-initiated if a different buyer wants the same or a similar special order in the future.
Thus, there is a strong need for a procurement system that takes into account special requisitions and resulting special orders, and uses the information related to such special orders so that the effort spent on fulfilling each order is not lost. In particular, it would be highly desirable to be able to receive a special requisition, and use a methodology to automatically select most likely suppliers without human intervention. If such an automatic selection process were available, then the special requisition could automatically be sent to each vendor for review and quoting. Moreover, it would be highly desirable to be able to receive the information from each supplier in a consistent format so that review of the received information could be automated as well according to pre-determined criteria to select the best supplier(s) for a possible special order. Additionally, it would be desirable to use the special expertise of both buyers and suppliers to create updated rules and item information associated with the special order. Therefore, future orders of the same item would not require the same special handling.
An inventive procurement system according to the present invention takes advantage of the information transferred between a buyer, a procurement fulfillment organization, hereinafter called a back office, and a supplier to automate and continuously update a catalog database of items.
In practice, a buyer searches for an item in one or more catalog databases using a parametric or text search strategy. If the item is not found, the user then creates a special requisition that will uniquely identify the desired item. The requisition is transmitted to the back office, which then forwards it to potential suppliers. One or more of the suppliers review and revise the requisition and send it back to the back office. The back office uses predetermined criteria to select one or more of the suppliers and then provides the buyer with the supplier information. The buyer may place an order for the special item.
A key advantage of the present invention is that both the catalog database as well as the special requisition are normalized using predetermined rules related to class, attribute, and value relationships that are already known to the back office, and must be followed by the buyer to create the special requisition. Generally, these rules are stored in a knowledge base. When the back office receives free form item data, the free form data is processed through the knowledge base to create the normalized database that a user searches. The rules used to create the normalized database are made available to the buyer to locate a specific item.
Thus, when making a special requisition for an item not in the database, the buyer first identifies the item using all available class, attribute, and value relationships used to create the normalized database. The remaining class, attribute and value relationship information required to uniquely identify the desired special item are then suggested by the buyer and forwarded to the back office.
The back office uses the preexisting information identified by the buyer to pre-select potential suppliers that are already associated with the selected classes, attributes, and values and automatically forwards the special requisition to them. Thus, no human intervention is required.
One or more of the suppliers review and revise the special requisition using the same relationship approach as followed by the buyer and dictated by the pre-existing rules of the knowledge base and then returns it to the back office. Thus, at least two experts (i.e., the buyer and the supplier) have proposed a revised normalization relationship to uniquely identify a desired item in accordance with the pre-determined rules. The revised normalization relationship may be used to create a new rule for the knowledge base to process future items with no further human intervention.
Moreover, by having a normalized item selection process, the desired item and related items may be readily added to a catalog database for future selection by other buyers. Such an approach takes advantage of the expertise of the buyer and supplier in fulfilling a special requisition while eliminating the need to undertake the same process for other buyers. The catalog database is automatically updated as required over time, generating catalog database updating in a real time fashion.
Yet another advantage of having a normalized approach to the special requisition process is to permit easy item comparison by a buyer. When more than one potential supplier of a desired item is located, having the item information in a normalized fashion according to class, variable and value relationships provides easy comparison of the various items by the buyer. Thus, comparison-shopping is expedited to the benefit of all parties since fewer returns are likely when the item criteria are clearly understood.
The features and inventive aspects of the present invention will become more apparent upon reading the following detailed description, claims, and drawings, of which the following is a brief description:
A procurement system 20 is illustrated in
Procurement system 20 is illustrated in greater detail in
If the desired item is found in the hot list as shown at decision point 52, then the user is next asked if there are any additional items to be ordered at decision point 54. If yes, then the item selected is temporarily stored at point 56 and buyer 22 returns to start point 42 to select the next desired item. If no additional items are desired then an order is created at point 58. Even if an item comes from more than one source such as a hot list or one or more databases 32, point 56 represents a common holding point such as a shopping cart. The option to modify the item selection is permitted at decision point 60. If an item is to be modified, then buyer 22 is returned to start point 42. Otherwise, any additional information is entered as required at point 62 and the order is submitted at point 64. In many cases, specific approval is not required if the standard buyer catalog database 32 or a hot list 44 is used. In some cases, however, an approval process is required by the buyer's organization as shown at point 66. Once the order has been approved, it is then placed by the buyer's organization as shown at point 68. Back office 24 informs one or more suppliers 26 who fulfill the order as shown by item 36, and discussed with respect to
If an item is not found using the hot list at decision point 52, additional methods for locating a desired item are available. First, the user is returned to start 42. If the hot list option is not selected, then at decision point 70 buyer 22 is asked if she will be using a standard buyer catalog database 32. If yes, then the standard database 32 is selected at point 71. Typically, a buyer accesses a standard buyer catalog database 32 that has been put together by his organization for routine purchases and that represents the purchasing preferences of the organization. If not, then additional database catalog(s) may be used as shown at point 72. One possible database is a cumulative global database representing all items tracked by back office 24 for many suppliers and buyer organizations. Other databases 32 may include an intermediate catalog based on geography, subject matter or the like. For example, an organization may have a regional database 32 including items available in a specific geographic region that is searched before a master catalog acting as a cumulative global database 32 is searched. Different databases 32 may be available depending on the characteristics of the buyer 22 including company, geographic information, and even buyer classification within an organization.
Once the desired database 32 has been selected, buyer 22 may conduct an advanced text search for a desired item as shown at decision point 74 and illustrated in
More typically, however, a buyer 22 takes advantage of the normalization features of system 20 by navigating the organizational hierarchy of items stored within one or more databases 32 and uniquely identified using a class/attribute/value relationship rather than conducting an advanced search. Thus, if a text search is not conducted as shown at decision point 74, buyer 22 then selects item classes as shown at point 90. Buyer 22 drills down through the various item classes that are presented in a hierarchical format. As shown in
Once a leaf class is selected, a parametric search engine is presented as shown at point 94. The parametric search engine presents the attributes associated with a specific item leaf class and valid values associated with each attribute. The attributes can be presented in a variety of selection objects, such as drop-down boxes, list boxes, and sets of check boxes as shown in the example of
Typically there are three types of attributes: static, differentiating, and dynamic. A Stock Keeping Unit (“SKU”) represents a specific item as defined by its differentiating attributes. For example, if an item is a shirt with differentiating attributes color and size, then blue XL shirt is a SKU. Static attributes define an intrinsic property of a product and do not vary based on a SKU. A differentiating attribute is an attribute that defines the uniqueness of a SKU. A dynamic attribute is an attribute that is associated with a product dynamically at buy time.
Once buyer 22 has selected all available attribute values then the buyer must determine if additional specifying is required as shown at decision point 96. If no additional specifying is required then system 20 passes to decision point 98. At decision point 98 the use of the class/attribute/value approach permits a comparison of multiple items that meet the parametric search criteria and the selection of the item that meets the buyer's specific requirements as best illustrated in the example of
While not illustrated in the Figures, item comparison may also be implemented when doing an advanced text search. By using the class/attribute/value approach, a consistent and specific determination of item elements is accomplished. In practice such an approach reduces buyer confusion regarding item features and provides a listing of all values available with respect to a specific attribute of interest to a specific buyer.
As shown in
It is also possible to compare items at point 100 using a graphical approach. As shown in
A preferred embodiment of the invention permits the use of a text search option even when undertaking a parametric search. For example, a user may desire to quickly find out if a database includes blue pencils. Therefore, once a class for pencils is selected, a text search may be desired to determine if there are any pencils with the color blue, thereby bypassing the rest of the class, attribute and value determinations.
To help a user determine the best balance between a parametric search and a text search, in a preferred embodiment of the invention additional information is provided as either a parametric search or a text search is undertaken. For a parametric search, under each class, the total number of children classes as well as the total number of items associated with the selected class is provided dynamically as part of an interactive feedback mechanism. As a user gets closer to a leaf class, the number of additional classes and the number of items associated with that class is reduced. For a text search, feedback in terms of the number of items satisfying the search is provided interactively and dynamically as the buyer 22 types a search string at point 78 or enters various operators at point 80 or parameter fields at point 82. When a text search is undertaken as part of a parametric search, the number of items is based on the search parameters already satisfied with respect to the parametric search Besides helping to further fine-tune a search strategy, a user is dynamically informed of problems including a mistyping or the use of unduly limiting Boolean operators.
Then system 20 continues as discussed above at decision point 52 where the item may be selected for purchase by clicking the “Add” icon and the process continues as discussed above until the procurement process is complete. If an item is not located using the class/attribute/value relationship approach then the user is returned to start 42. As with the advance text search option, if the select class option is again selected the most recent leaf category will be provided to avoid the need to re-enter all of the search criteria
As noted above, it is possible to select from a variety of catalog databases 32 at point 72. If an item is not located during an initial search of previously selected databases 32, requiring a return back to start point 42, system 20 may be designed to automatically add pre-determined additional databases. In the alternative, a buyer 22 may choose which databases 32 to search. In either case, the addition of different databases 32 after an initial search of one or more databases 32 is an example of a cascading search.
To facilitate such a search, particularly when conducting a parametric search, in a preferred embodiment of the invention the various databases 32 include consistent classese, leaf classes, attributes, and values. Thus, information provided in an earlier search is passed to a later search, including the class node or path, attributes for the class, and the corresponding values for the attributes. For example, assume a user is looking for a blue pencil, but the usual database 32 does not include such a pencil. Moreover, assume that the first database searched includes an attribute called “Pencil Color”, but lacks the value of blue. A different database 32 includes a pencil with an of the attributes and values located with respect to the first database 32, but includes the desired blue pencil. If that database is searched then the value “blue” will appear under the attribute called “Pencil Color” with the organization of the classes, attributes, and attribute values preferably the same. Thus, a user will only have to look under the appropriate attribute listing for “Pencil Color” since the search parameters will already be populated based on the earlier search.
To populate a pre-existing search it is generally preferable to send the search criteria between databases 32 as opposed to bringing an entire catalog itself down to the location of a buyer 22. The search criteria are either sent to one or more different servers or possibly to one or more additional databases 32 on the same server as that of the database 32 originally searched. The population of the search criteria can be done in parallel simultaneously against multiple catalogs as represented by databases 32 to provide initial feedback to a user concerning the number of items that meet the search criteria for each of the available databases 32 as discussed above. When a buyer 22 selects particular catalogs or databases 32 to browse, that buyer is then typically presented with the parametric search screen for those products that match the search criteria found in those catalogs.
While the present invention recognizes that protocols such as Extensible Markup Language (“XML”) may be used to populate a search criteria, when a typical Internet web browser such as Internet Explorer or Netscape Communicator are being used, the search criteria are included in a Uniform Resource Locator (“URL”) query string.
A cascading search using a common hierarchy provides a number of advantages. For example, it makes it easier to copy search information when searching between databases 32 without requiring undue additional re-inputting from a buyer 22. Additionally, it permits synchronization between databases 32. It is easy to move items between databases 32 when they have a common hierarchical scheme. Moreover, a common interface may be provided to a user even when searching between different catalogs. For example, if search results are provided in a top frame of a web browser for a database 32, the same frame and output format may be used for a different database 32, where information about the second database may or may not be given. Thus, movement between databases 32 may be generally transparent to a buyer 22.
While it is desired to have a common hierarchy scheme between databases 32, system 20 recognizes that it may be desirable to have exception lists between two different databases 32. For example, an exception list may be used to provide translations between different languages. Thus, a class, attribute or value in one language will be compared with a translation table to permit it to be matched with a corresponding class, attribute or value in a second language associated with a different database 32.
If a user desires additional specifying at decision point 96 of
If a leaf class has already been selected or once a leaf class is determined, then a modified parametric search screen with the most recently selected acceptable class is provided at point 108. It may also include the pre-selected attribute and even value information. Similarly, unlike the regular item selection process discussed above, the attributes and value selections are not restricted to the database 32 originally selected by the buyer 22. Instead, they include all possible attributes and values for a selected leaf class in all catalog databases 32 available from back office 24. Thus, in many cases, a user will not have to manually enter any additional information into the modified parametric search screen. However, unlike the screen shown in
An example of a structured requisition form is illustrated in
Once the structured requisition is completed at point 114, it is submitted to back office 24 by way of communications link 28. Communications link 28 was discussed with respect to
Upon receipt of the structured requisition at point 118, back office 24 reviews the requisition to determine if it is correct as shown at decision point 120. If not, then buyer 22 is notified at point 122 by way of link 28 and the structured requisition subsystem 102 terminates at end point 124.
If the requisition is correct, then it is compared with all items in the master database 32 corresponding to the selected classes/attributes/values as shown at decision point 126. If one or more matches are found at decision point 126, communications link 28 provides the purchase information to buyer 22 for order submission at point 128. In a preferred embodiment, an electronic mail message is sent to buyer 22 with a hyperlink that will be recognized by a web browser such as Netscape® Navigator or Microsoft® Explorer. Buyer 22 clicks on the hyperlink and is routed to a screen very similar to the comparison screen of
A key advantage of the present invention occurs if the desired item is not in the master catalog database 32. At point 132 the structured requisition is automatically reviewed by back office 24 using the class/attribute/value information and normalized database information at its disposal. Based on the information provided by the buyer 22, automated systems of back office 24 route the structured requisition to potential suppliers as shown at point 134 using communications link 34. Communications link 34 was first discussed with respect to
In a preferred embodiment, an electronic mail message is sent to potential suppliers again using hyperlink and web browser technology, as discussed above. Preferably, as discussed further below with respect to
Typically, all suppliers 26 receiving a structured requisition are given a predetermined amount of time to provide a revised requisition. When either the time for replying expires or all pre-selected suppliers provide a revised requisition, then at point 142 back office 24 compares the revised requisitions to determine the best supplier or suppliers for the specific buyer 22. Back office 24 uses a wide range of objective and subjective criteria to determine the best supplier(s) including geographic location, price, reputation, timeliness, and the like. Often, the criteria include those predetermined by the buying organization based on its specific requirements or desires. As a result, the wishes of the buying organization are automatically considered to expedite the purchase process and reduce both cost and inconvenience.
Once one or more items are screened as best meeting the needs of buyer 22, they are then provided to the buyer as shown at point 128, and discussed above.
As shown in
The use of a normalized parametric search has been discussed above. Every item in a catalog database is normalized in terms of class hierarchy with a final leaf class being followed by a series of attributes and attribute values. As a result of the normalization process, item determination and searching is greatly enhanced. When undertaking a structured requisition, the same normalized data is used to the extent that it is available to buyer 22. A specialized supplier having expertise with respect to the class/attributes/values with which a requisitioned item is associated then further refines the structured requisition. Thus, both a knowledgeable buyer 22 and a knowledgeable supplier 26 are providing normalized relationships between an item and its defining characteristics. The relationships are used to provide selective rule updates to the common language generator knowledge base 30 and to an associated catalog database 32 using the predetermined rules of the knowledge base, as updated from time to time, to convert raw or free form data into the preferred normalized format. Thus, once a catalog database 32 is established, the catalog database is continuously fine-tuned by suppliers, buyers, and changing market conditions of which both are aware.
Items may be moved or copied between different databases 32 based on various business requirements such as purchase history as tracked by back office 24. The information is easily moved by using the common parametric hierarchy disclosed in the present invention.
While a separate knowledge base 30 is illustrated, it is possible to bypass the knowledge base 30 completely and update a database 32 directly, so long as the class/attribute/value relationships are maintained. An advantage of using a separate knowledge base 30, however, is where one or more new rules are specified that are later available to be able to normalize free form data.
As shown in
Catalog database 32 has been designated fairly generically. The database 32 may be a separate database of items from a specific supplier. Alternatively it may be a composite database. When acting as a composite database, it may represent items from multiple suppliers that can be organized for and based on the buying habits of a single buying organization (e.g., the standard buyer catalog database discussed above) or upon the buying habits of a group of related buying organizations (e.g., as in 1s hospitals and doctors' offices that are members of a group purchasing organization). Therefore, when a database 32 is updated, it may be any one or a combination of databases 32 depending on the desires of the supplier buyer 22 or her buying organization, suppliers 26, and back office 24.
At decision point 158 if the decision is made that the normalized information associated with the revised requisition is new, then the determination must be made whether to revise the common language generated knowledge base at decision point 172. If the decision is made not to update the knowledge base then sub-system ends at termination point 166. On the other hand, if the decision is made to update the knowledge base then it is updated with the new information at point 174. Then the decision must be made at point 176 whether to update existing catalog entries based on the updated knowledge base. If yes, then the entries are updated at point 178 using the updated normalized information that can be associated with the raw or free form item data associated with each entry. If no, then the entries are not updated. Then subsystem moves on to decision point 162 as discussed above.
The system based interaction between buyer 22, back office 24, and supplier 26 is explained in greater detail in
After buyer 22 logs on, item selection is accomplished using a hot list, advanced text searching, class/attribute/value selection, or a structured requisition request as discussed above. The item selection process requires that a catalog database 32 be queried for information directly by buyer 22 as shown by line 204. An order submission is made as shown by line 64 and discussed above, that is passed to application 40 from buyer 22. Any required approval process is shown by item 66 and an order placement made after any approvals is shown by item 68, both of which were discussed above. Once an order is placed, order status may be passed from interface 202 to application 40 and then to buyer 22 as shown by line 206 until the order is fulfilled by a supplier 26.
If a structured requisition is required, the complete listing of available classes/attributes/values is retrieved from the master or global catalog database 32 as shown by line 208. The structured requisition is passed from buyer 22 to application 40 by line 210 where the structured requisition is submitted as shown at point 114, much as if it were an order. Alternatively, the information may be passed directly from buyer 22 to back office 24 by way of communications link 28. If one or more items are located using structured requisition subsystem 102, the information regarding item availability is typically passed from back office communications interface 202 to application 40. Next, the information is passed to buyer 22 in the form of an electronic message with a link pointing buyer 22 to a server with the desired item information. The hyperlinked information may be within back office 24 or transferred directly to application 40 as desired.
As shown in
As shown in
In general, interface 202 is used to communicate with buyer 22 by way of application 40 as noted above. However, when accessing catalog databases 32, buyer 22 can preferably bypass buyer interface application 40 as long as authentication has been completed. Interface 202 is also used to communicate with supplier 26. As shown by line 216 all order placement, acceptance, and order status information is shared between interface 202 and supplier 26. When structured requisition subsystem 102 is being used, interface 202 provides the structured requisition as shown by line 218 and receives a revised requisition as shown by line 220. Supplier item data is shown by line 221. Each of these lines comprises a portion of communications link 34 shown in
Supplier 26 has only limited contact with buyer 22 as shown in both
Interface 202 uses information it receives from both buyer 22 and supplier 26 to selectively update catalog databases 32 as discussed with respect to catalog update subsystem 150 in
Knowledge base 30 is the glue that relates an item selection process with the items actually stored within a database 32. It is formed using pre-determined rules that have been established over time that relate items to class/attribute/value characteristics or relationships. These pre-determined relationships are pulled from CLG knowledge base 30 as shown by line 222 to be compared with potential classes/attributes/values received from supplier 26 as shown by line 220. Selective updating of the rules governing knowledge base 30 is shown by line 224. Line 226 represents the free form supplier item data to be normalized and then loaded into a database 32. The data is normalized using the class/attribute/value relationships that govern the common language generator and knowledge base 30. In some cases, however, a supplier 26 may be able to update its own database 32 without having to go through the Common Language Generator, particularly if an appropriate parametric structure has been set up using classes, attributes and values. For example, a supplier 26 may add or remove attributes or values associated with particular items or correct mistakes in attribute values and product descriptions. It may also delete or add entire items or groups of items. Under some circumstances a buyer 22 or the buyer's organization may be permitted to update its own local database 32 in a similar manner without direct involvement of back office 24 other than by providing the necessary interfaces.
Once the supplier data has been normalized using the class/attribute/value relationships of the CLG knowledge base 30, it is sent to an appropriate database 32 as shown by line 230 where the database entries are either imported or updated as shown by line 232. The complete listing of all class/attribute/value relationships used by CLG knowledge base 30 are also contained within databases 32 and transferred from the knowledge base as shown by line 234. These relationships are used to undertake parametric searching and item specifying as discussed above. The class/attribute/value relationships are then updated when required as discussed above with respect to
The disclosed embodiments and examples are given to illustrate the present invention. However, they are not meant to limit the scope and spirit of the present invention. Therefore, the present invention should be limited only by the appended claims.